That is what happens when the input is not isolated properly from the control circuitry. For those that do not know what the problem is, thermocouples put out millivolts that has to be measured on a voltage bridge to get an accurate reading. Then do calculations to determine exact temperature as the temperature does not correspond to a linear output it is logarithmic.
I've really missed out then, I wouldn't say we've had a very good May where I am (SE UK) other than a handful of days. It was good to have a dry bank holiday for a change though.
I'm not usually into heat, but today was outstanding. The TT races are on at the moment and it was the Ramsey Sprint today. I enjoyed wandering about in the sun.
I had something similar happen with a circuit I was testing, where switching noise would couple through the thermistor's leads and get into the cooling fan control circuit. The fan would instantly stop rotating and the circuit, overheat.
Couple of possibilities: 1. This is exactly what will happens when JavaScript/PHP developer goes into electronics. Slap bunch of libraries together and push to production. 2. Student's first prototype fished out from EasyEDA and pushed directly into production. Possible by JS/PHP programmer, because testing not for programmers, it's for users in production.
I've seen all the video. He shorted the terminals "to see what happend", and the reading was a very low value the first time (6º). Then tried again and it red 28º what is more reasonable but he sed it's "odd". It's not odd for me (a little high but not "odd"). It is normal. The previous value of 6º was odd. A new test after changing temperatures red 22º for a moment and he sed it is "strange". It is not strange at all. It es a correct reading when you short-circuit the terminals.
Alejandro Vargas. What Clive should have done (for his test) was short the + and - terminals (with a short piece of wire) but not hold on to it with his hands. His hand creates a short and/or picked up noise. Then, in theory, the output of the 6675 should represent the temperature of the cold junction. In my experience, it is best to also use "digital isolation IC" to prevent corrupted signals through the SPI bus, especially if you use this type of metal shielded thermocouples to sense the temperature of metallic parts (That should themselves be grounded)
John Coops. Alejandro did *not* suggest to short the terminals and Clive got *mixed* results. We were just saying, the expected result is ambient temperature, which seems to me, none of Clive's attempt gave.
You are probably not far off the mark that it's a kid's first design. Or a design from a graduate with no experience. Guy in China decides to set up a company making stuff like this, hires a new graduate on the cheap (or asks the hobbyist kid of a friend) to do the design, and away you go. After the first production run he spots all the problems and flogs them off cheap because they're crap. And that's one of the better companies. The bad ones never realize the product is crap and keep making it. The *really* bad ones realize it's crap but it sells, so they keep making it because there's no comeback.
Shorting a thermocouple input will usually read the temperature of the cold junction compensation sensor (resistor) (the temperature at the terminal strip). I wonder if this even has cold junction compensation.
But it should be mounted as close to the connection terminals as possible to correct the right junction. Also, it is recommended to always use common mode filtering.
Yeah does have cold junction but I doubt this is very accurate in this implementation, a terminal strip like this is one of the worse connection methods for a thermocouple as it most likely introduces more dissimilar materials and thus more parasitic thermoplastic, the main sensor disconnect/connector should maintain the same material for metal A and B right up to metal C, metal C being copper bolts, copper nuts, copper spades, copper legs and copper PCB tracks.
But those junctions are in symmetrical pairs. Also there is no copper to copper contact between... basically anything. When you solder an IC on a PCB, the IC legs are made of steel, not copper, and even if they were made from copper, there's a thin layer of tin in between! No matter how thin, a junction is a junction. And the IC legs are connected to the IC substrate with gold wires.
Hi Clive, As Alejandro Vargas noted, this is a k-type probe, that is, it senses the temperature difference between probe tip and the MAX6675 side. The MAX6675 than compares the reading with a virtual 0° reference point by reading the ambient temperature itself. As a result, the MAX shold be placed as close as possibe to the "cold side" of the probe (the terminal block) far from heat-generating parts.
The cold junction, and the Max, are all at ambient temperature. Theoretically you are correct, but the impact on the video is small. There are many other more likely problem sources mentioned in the comments.
One thought comes to mind: EAGLE, and many similar packages have reusable circuit modules you can just drop into a design. If this were a student's design from scratch I would have expected the component placement to be further apart, but the fact the various modules have components clustered up might mean that the modules were pulled out of libraries and placed down.
I needed help on a W1209 temperature controller and your old video was right on the button. Odd I had missed it, only found it by accident. You should have mentioned it on this video. I have just ordered 2 more at 99P each on EBAY. It's got to be the best value for money ever. A little fiddly but once you know what does what it's superb.
Nice video! One suggestion: You could improve the readability of the LED display on camera by placing a small bit of red plastic foil on top of it - I think Julian Ilett has done that successfully.
Remember a thermocouple is a differential sensor with a sensing junction and somewhere a reference junction (where the special termocouple wires connect to conventional wires). So when you short the thermocouple input you expect the reading to go to room temperature, (the temperature it thinks the reference junction is at). When the short took it down to a low reading was the board colder by any chance?
Most temperatue controllers are set for up scale burn out i.e. open circuit sensor drives the display up scale which would normally turn off the load by passing the upper set point, shorted input should read ambient temperature i.e. cold junction temperature for a thermocouple sensor.
This may seem a bit off topic but you think you could do a video on tesla coils? They're so complicated and there aren't many well made videos explaining them. I could see a nice 40 min video of the ins and outs of tesla coils being a huge massive help to the community of electronics, spark gap, slayer exciter, sstc, drsstc, vttc, etc
An open thermocouple often makes the controller read high, it's called 'offscale high' and is designed to force the relay off. The screen of the thermocouple will be earthed at the thing it's screwed into, or else you can earth it at the controller if the psu is isolated as it should be. Leakage from a heating element may also interfere with thermocouples so again the two should be bonded together and to earth. Shorting the thermocouple out should make the display read room temperature because it should show the cold junction compensation value (should be measured locally on the board by a diode or thermistor).
My experience is that of BitBastelei. When i would submit a preliminary version of university project and list its known/suspected bugs like "lack of debouncing" etc, i had a few days left and i'd still fix them up, they were like... look it passes simulation, let it go, it's good enough, we really just care that you brought in the concepts that we touched upon in the lectures, it's not an actual product that anybody is going to make, it doesn't have to actually work, YOU'RE GETTING TOP MARKS ANYWAY LET IT GO OK PLEASE?
no high school student has the time, the vast knowledge, money or experience to design a micro-controlled device at even the Chinese bad engineering level and make a double sided PCB for it, the surfacing mount soldering it together, there is just too much to learn and know, to get to that level, this would be an A grade uni engineering student efforts. The best is home made PCB from the hobby store and a off the shelf microcomputer they can program.
@bjtaudio I designed and programmed a similar project back in high school in the 90s. The school didn't have facilities for manufacturing multi-layer boards so I built it single-layer and submitted the multi-layer design. That was an A. If I'd submitted something like this during my engineering degree it would not have got a passing grade. Later, as a postgrad teacher, if this had been submitted by a first-year student I'd have declined to mark it, but instead have invited the student to review their design process with me. Engineering standards are a little higher than you seem to believe :D
i guess it's kinda indicative of how readily available pcb printing and other related technology is over there- even people with basically no idea what they're doing can create circuits / products. it's cool in a way, while also scarey.
First idea, when I saw it was, that the temperature sensor was wrong polarized. Trying a potentiometer instead of the sensor will perhaps bring a little light to what exact the circuit-board is doing. Sorry for bad English.I'm from Germany.
Connecting a potentiometer in place of K-type thermocouple does nothing, as the thermocouple has no resistance whatsoever, it generates a little voltage instead.
Monkeh, if you want to electrically measure the wire length, you may ponder the resistance of a K-type thermocouple. If you want to measure temperature, as intended, the deciding parameter in a thermistor is its resistance, and as opposed to that, in a thermocouple, it's the induced voltage. And yeah it has no resistance whatsoever. K-type wire often has less resistance than the sloppy Chinese screw terminal it connects through.
Why is the left-most digit flickering a lot more than the others? At first I thought it might be some sort of synchronized rolling shutter effect (temporal aliasing) with the camera, but it did the exact same thing after he power cycled the board as well. If the entire display is being driven from the same chip, and even using the same single row, the only thing that I can think of is that there's something going on with the column driver corresponding to the leftmost digit.
I'm pretty sure that's happening, because the pins are shared for driving the 7-segment displays and reading buttons. One of the digits has to be switched to "input". Since a dedicated driver/button reader shouldn't be that bad and almost fail at what it's supposed to do (I mean, giving a useable display output is it's purpose), my guess is that whoever wrote the program read the buttons, but didn't set it back to output right after that.
superdau I don't think the TM1650 allows the microcontroller to set the I/O state of the 7-segment pins directly. When reading which key is pressed, the microcontroller just sends a byte to the SDA/SCL pins and then reads a byte back. The TM1650 is in control of the multiplexing between keypad polling and 7segment led illumination.
Yes, Clive, you performed a fair test of this module. I would be ashamed to offer this "product" for sale when a minute or two of cursory testing shows it to be so flawed and in need of further development. Then again, I'm not Chinese.
Those LED segment displays are hard to see on camera. I figured out a little tip for that, especially the bright ones like the blue ones. When you peel off that protective shield it sometimes makes it brighter. Take a little strip of thin Kapton tape and place it in the segment display and it makes a big difference in cameras. Makes a huge difference if it’s a really bright one and easier to read on camera and in person. Hole this helps.
I’ve seen that non-shorting issue in a few cheap terminals... i think the screw head isn’t part of the electrical design.. only the actual ‘plates’ that sandwich the terminal/spade. If you’re unlucky, and the screw thread doesn’t touch the actual wire or spade - then it simply passes through the conductors to the captive nut on the other side of the mechanical assembly. I can always get a connection at the mating surfaces... Odd, but there may have been a reason ??
LOL, I ordered a Temp control from BangGood just weeks ago. I had to insure this was not the one I had. I've used the MAX6675 with Arduino before. A Cap across the K ThermoCouple leads helps with noise. The MAX 6675 is a Current Amplifier. The K ThermoCouple generates a Current . They are immune to wire length and voltage loss.
2:45 my guess: they used whatever their board design tool had for the standard pad size and just didn't think it through. i'm also kinda scared by the size of those pins, wouldn't use that terminal for any high power application to begin with, i feel like you probably want a bit larger diameter pins there. even something "low performance" like most barrel sockets have giant pins compared to that terminal strip.
The MCU is a modern variation of the classic 8051 architecture. It might be worth it just as a devkit (if you have the necesary programmer. Bootleg ones go for $2 on ali.)
There is technically 3 connections to that thermal couple... if you look at the connections they are not part of the shielded lead, that shielded part needs to be grounded to the ground on the device. This is normally done when its attached to what ever your measuring via the chassi So if you add a wire to the shield and direct that to device ground, it will stop swinging
Another key thing is that thermocouples physically measure the difference between the temperature at the two wire ends, and the application needs to control the board end reference temperature to make sense. Scale may be in Fahrenheit, with a wrong reference for 32°F / °C, thus interpreting almost no difference as slightly below 30° nonsense.
2:39 I think the answer falls in line with your later conclusion that this is a sort of "text book" circuit design but with *many* errors. A sizable relay was used given the application, it was well bonded to the board, but the designer(s) utterly failed to realize (or care about?) the simple fact that the large loads this thing is supposed to be switching will be attached to those tiny terminals causing an obvious failure point. Only a hobbyist but that seems to make the most sense to me.
K type thermocouples often use Red as the negative output.... and often not isolated from the braid and mounting fixture. This may explain the 'noise' effect when handling and down register when the couple is heated. Yeah... creep distance / clearance on the power supply and relay tracks is an issue... fix with cutting board between the tracks to effect air gap rather than tracking across a flat surface. BUT, you shouldn't need to do that.
Not sure if the testing methodology is correct for a thermocouple. For a PTC thermistor it would be but really if you shorted out a thermocouple you'd think it'd just display the ambient temperature. Because the connection to the thermocouple wire is in fact another junction at ambient temperature. Disconnecting the thermocouple just makes it read whatever the input is at if you leave it floating.
Undefined Lastname do you mean the shield is connected to the blue connector (IN -)? With my comment I meant a permanent connection from DC ground to the shield. But maybe that is already the same.
That thermocouple negative pin should be grounded RIGHT at the chip, instead they routed it threw the serial port, the voltages that thermocouples work on this will A) never hit 0, and B) will be very noisey EDIT: it also looks like they intended a calibration via a pot to the analog pins but never fitted it
Remember this temperature controller is modular, it's very modular, it's very very modular and it's a modular design that even the Borg would not assimilate into there collective even as a adjunct to there uni matrix zero one.
if you short the thermocouple terminals, I'd think it should display the ambient room temperature, as it would simulate 0 volts from the sensor. the way thermocouples work, they detect the temperature difference between the sensing end and the end connected to the terminals. there's normally a second temperature sensor on or near the board (preferably thermally coupled to the terminals) which detects the ambient temperature, then uses the thermocouple probe for fine adjustment. if the probe end is warmer than the terminal end of the thermocouple, it'll feed a small positive voltage into the board. if the probe end is cooler than the terminal end, it will feed a small negative voltage into the board. it actually reads a relative temperature, not absolute.
I love watching these videos. Though more and more I'm realizing that Clive has probably forgotten more about electronics than I'll ever know in my life time. Keep the videos coming! Cheers.
Questions>>>> Wondering out of curiosity... One- The coupling of your hand on therm couple. Try a say 20 to 100 pico Farad cap across coupler input? Also wondering if put say 220 to 470 micro Farad across power input? See if that fix that issue. Far as some of the other things, relay going into bipolar disorder, no error on disconnect of probe so temp runs amok, these things I am laughing at myself. But was courious bout change in temp on that and if a few pico's on the them might help lots. Like the videos.
Well, I've bought this for my DIY CNC Router to 3D Printer conversion. It was cheap and I thought it might work for some time. We'll see :D It's arrived, but there are other components to come before I can start building the printer module.
No investigation as to WHY there's interference on the thermocouple input? The Max6675 datasheet says that the thermocouple negative should be grounded to the chip. Is it?
dear sir i have bought a 12volt crock pot from road pro worked great for about a month then i tried it didn't work the light came on no heat ruined my dinner.so i bought another one now the same thing lasted 2 weeks ruined another dinner contacted amazon they just said you bought it .i am not buying another i took it apart the wire is broken so i was just wandering what type of wire it goes to the element.it is a solid wire.thanks for your videos
Could it be the that one of the sides of the thermocouple chip isn't grounded? Since the example schematic shows that one of the terminals should be grounded - which also explains how it starts reading about right when you ground the probe with your hand.
If you were so inclined you could connect the microcontroller to a suitable reader and download the code. Then you could attempt to correct the omissions and re upload the code and see if it functions any better.
Hi, Clive have you checked that the thermocouple is connected the correct way round? I didn't see you actually heat/cool the thermocouple, just play with it noise characteristic. Have you tried a 0.1uF capacitor across the input terminals and one from one input terminal the gnd. The capacitor on the PCB near the controller and the thermocouple chip is connected to what? Tom..
Could you de-solder and remove all the on board power supply components (maybe pop in a 100nF and 100uF in parallel just for added noise removal )and just bridge over the input to the supply rails using your bench power supply set at the right voltage and see if the problem still persists? That would eliminate the on board power supply as a potential noise source? Could a suitable resistor across the probe input terminals also be used to stabilise/calibrate the probe? Use a different probe? Such a fun board with so many experiment possibilities :D
Hey Clive! In my University in Germany we have Spartan-3E FPGA Boards. We use them to design our very own 4bit microprocessor from the ground up. I just thought it might give an interesting Video idea for how it might be used on all sorts of different prototyping
if shorting the sensor does not make a difference, I really start to wonder if that sensor is doing anything. Wonder what happens if you first ground it and then short the sensor. Maybe reading some screw-wire-screw thermocouple. There is one thing good to say about this design though. I got the impression its very modular. Not sure what gave that away
Hey clive could you do a video where you muck about with some old 74xxx TTL logic chips? Binary counters, 7 segment displays, 555 timer.. I know a lot of that is redundant today with Arduinos and RaspRi's.. but it's still good stuff.
I think the pic has a resolution of 1024 bits. You just get a decimal readout of the anologue sensor's reading. Shame about the lack of ground / bad program.
I really like the relay disco mode, because it's such a stupid mistake anybody with any programming experience would have caught immediately. If relayOff and temperature < TurnOnTemp then relayOn; If relayOn and temperature > TurnOffTemp then relayOff;
I have a different but similar module. But my p1, p2, etc. are not different presets but are functions such as number of degrees hysteresis. Below is another module different again. Are you sure of the functions on your one? P0 Heat C/H C P1 Backlash Set 0.1-15 2 P2 Upper Limit 110 110 P3 Lower Limit -50 -50 P4 Correction -7.0 ~ 7.0 0 P5 Delay Start Time 0-10 mins 0 P6 High Temperature Alarm 0-110 OFF
The take away for me, the sensor connector itself has a bad connection to the board, when you pushed down with the scissors it forced a connection, I'd say it needs to have the solder run.
You should be able to get that to read right. I've found that thermocouple are quite good antennas, I think the dissimilar metals rectify noise and create an offset. You should be able to sort that
If the thermocouple has a different range than the unit is calibrated to, you'll see problems like you're seeing. Also, did you try running a line from the probe- to ground? while the probe is connected that is?
Shouldn't the two thermocouple leads be taken to the input independently of the screen braid which should be grounded. Even if one of the thermocouple leads is grounded at the chip? Also could it be something as simple as a faulty decoupling capacitor? How about another video improving the design?
So your body temperature is 17C. No wonder you never bother with central heating.
Jim Fortune
Technically, Clive is dead!
Clive confirmed for being a vampire.
Clive has played a herd member in The Walking Dead and was the only extra that didn't have to act.
Clive's environment is cold because he only heats from within with the handy baby bath water heater XL. :)
Maybe that's why he was worried about reflections from the mirror base of that power bank in the previous video!
That is what happens when the input is not isolated properly from the control circuitry. For those that do not know what the problem is, thermocouples put out millivolts that has to be measured on a voltage bridge to get an accurate reading. Then do calculations to determine exact temperature as the temperature does not correspond to a linear output it is logarithmic.
The most surprising thing about this video is that your workshop is as warm as 20 degrees.
UK's just had the hottest May since records began...
I've really missed out then, I wouldn't say we've had a very good May where I am (SE UK) other than a handful of days. It was good to have a dry bank holiday for a change though.
I'm not usually into heat, but today was outstanding. The TT races are on at the moment and it was the Ramsey Sprint today. I enjoyed wandering about in the sun.
It’s not been bad today up here in the middle although Sunday was better.
23deg upstairs now with all the windows open.
Kind of sucks to be in Kent then, it's been under 20 degrees here today, a bit stuffy and in need of a thunder storm.
I had something similar happen with a circuit I was testing, where switching noise would couple through the thermistor's leads and get into the cooling fan control circuit. The fan would instantly stop rotating and the circuit, overheat.
Couple of possibilities:
1. This is exactly what will happens when JavaScript/PHP developer goes into electronics. Slap bunch of libraries together and push to production. 2. Student's first prototype fished out from EasyEDA and pushed directly into production. Possible by JS/PHP programmer, because testing not for programmers, it's for users in production.
If you short-circuit the thermocouple terminals, you will read the ambient temperature (that is measured by a thermometer in the converter chip)
John Coops. Try to say something constructive, like Alejandro tried to do.
I've seen all the video. He shorted the terminals "to see what happend", and the reading was a very low value the first time (6º). Then tried again and it red 28º what is more reasonable but he sed it's "odd". It's not odd for me (a little high but not "odd"). It is normal. The previous value of 6º was odd. A new test after changing temperatures red 22º for a moment and he sed it is "strange". It is not strange at all. It es a correct reading when you short-circuit the terminals.
Alejandro Vargas.
What Clive should have done (for his test) was short the + and - terminals (with a short piece of wire) but not hold on to it with his hands. His hand creates a short and/or picked up noise.
Then, in theory, the output of the 6675 should represent the temperature of the cold junction.
In my experience, it is best to also use "digital isolation IC" to prevent corrupted signals through the SPI bus, especially if you use this type of metal shielded thermocouples to sense the temperature of metallic parts (That should themselves be grounded)
John Coops. Alejandro did *not* suggest to short the terminals and Clive got *mixed* results.
We were just saying, the expected result is ambient temperature, which seems to me, none of Clive's attempt gave.
John Coops. The have medication for adult tantrums these days. Take a look 😀
You are probably not far off the mark that it's a kid's first design. Or a design from a graduate with no experience. Guy in China decides to set up a company making stuff like this, hires a new graduate on the cheap (or asks the hobbyist kid of a friend) to do the design, and away you go. After the first production run he spots all the problems and flogs them off cheap because they're crap.
And that's one of the better companies. The bad ones never realize the product is crap and keep making it. The *really* bad ones realize it's crap but it sells, so they keep making it because there's no comeback.
Impressed with the two videos in one day!
Still really interested in seeing your work space/camera set up :)
Shorting a thermocouple input will usually read the temperature of the cold junction compensation sensor (resistor) (the temperature at the terminal strip). I wonder if this even has cold junction compensation.
Of course it does. MAX6675 has cold end compensation built right in and always operational.
But it should be mounted as close to the connection terminals as possible to correct the right junction.
Also, it is recommended to always use common mode filtering.
Yeah does have cold junction but I doubt this is very accurate in this implementation, a terminal strip like this is one of the worse connection methods for a thermocouple as it most likely introduces more dissimilar materials and thus more parasitic thermoplastic, the main sensor disconnect/connector should maintain the same material for metal A and B right up to metal C, metal C being copper bolts, copper nuts, copper spades, copper legs and copper PCB tracks.
But those junctions are in symmetrical pairs. Also there is no copper to copper contact between... basically anything. When you solder an IC on a PCB, the IC legs are made of steel, not copper, and even if they were made from copper, there's a thin layer of tin in between! No matter how thin, a junction is a junction. And the IC legs are connected to the IC substrate with gold wires.
Hi Clive, As Alejandro Vargas
noted, this is a k-type probe, that is, it senses the temperature difference between probe tip and the MAX6675 side. The MAX6675 than compares the reading with a virtual 0° reference point by reading the ambient temperature itself. As a result, the MAX shold be placed as close as possibe to the "cold side" of the probe (the terminal block) far from heat-generating parts.
The cold junction, and the Max, are all at ambient temperature. Theoretically you are correct, but the impact on the video is small.
There are many other more likely problem sources mentioned in the comments.
Oscar Gr that's right. I perfectly agree.
One thought comes to mind: EAGLE, and many similar packages have reusable circuit modules you can just drop into a design. If this were a student's design from scratch I would have expected the component placement to be further apart, but the fact the various modules have components clustered up might mean that the modules were pulled out of libraries and placed down.
The tip of the probe looks like it's meant to be screwed into something earthed, and that the shielding braid gets grounded that way.
Keep in mind though that the threads on these thermocouples are usually M6*1,25 so you need a fairly special tap
jhsevs If the case is earthed, you don't need special taps.
You screw it in to mount the thermocouple. It's threaded
@@CyberlightFG You need to thread the hole that the tc goes into.
@@jhsevs on Ebay they are also available in M6x1 and M8x1.25
I needed help on a W1209 temperature controller and your old video was right on the button. Odd I had missed it, only found it by accident. You should have mentioned it on this video. I have just ordered 2 more at 99P each on EBAY. It's got to be the best value for money ever. A little fiddly but once you know what does what it's superb.
Nice video! One suggestion: You could improve the readability of the LED display on camera by placing a small bit of red plastic foil on top of it - I think Julian Ilett has done that successfully.
Remember a thermocouple is a differential sensor with a sensing junction and somewhere a reference junction (where the special termocouple wires connect to conventional wires). So when you short the thermocouple input you expect the reading to go to room temperature, (the temperature it thinks the reference junction is at). When the short took it down to a low reading was the board colder by any chance?
Most temperatue controllers are set for up scale burn out i.e. open circuit sensor drives the display up scale which would normally turn off the load by passing the upper set point, shorted input should read ambient temperature i.e. cold junction temperature for a thermocouple sensor.
Have you tried swapping the leads around for the themocouple? I have seen similar things happen when they are wired backwards.
This may seem a bit off topic but you think you could do a video on tesla coils? They're so complicated and there aren't many well made videos explaining them. I could see a nice 40 min video of the ins and outs of tesla coils being a huge massive help to the community of electronics, spark gap, slayer exciter, sstc, drsstc, vttc, etc
An open thermocouple often makes the controller read high, it's called 'offscale high' and is designed to force the relay off. The screen of the thermocouple will be earthed at the thing it's screwed into, or else you can earth it at the controller if the psu is isolated as it should be. Leakage from a heating element may also interfere with thermocouples so again the two should be bonded together and to earth.
Shorting the thermocouple out should make the display read room temperature because it should show the cold junction compensation value (should be measured locally on the board by a diode or thermistor).
Some Chinese high school student's science project. He probably got a B-.
A cheap product that doesn't work like it should, should have gotten an A.
Well, a lot of students wouldn't be able to design a board and software that at least is able to do something ;)
My experience is that of BitBastelei. When i would submit a preliminary version of university project and list its known/suspected bugs like "lack of debouncing" etc, i had a few days left and i'd still fix them up, they were like... look it passes simulation, let it go, it's good enough, we really just care that you brought in the concepts that we touched upon in the lectures, it's not an actual product that anybody is going to make, it doesn't have to actually work, YOU'RE GETTING TOP MARKS ANYWAY LET IT GO OK PLEASE?
no high school student has the time, the vast knowledge, money or experience to design a micro-controlled device at even the Chinese bad engineering level and make a double sided PCB for it, the surfacing mount soldering it together, there is just too much to learn and know, to get to that level, this would be an A grade uni engineering student efforts. The best is
home made PCB from the hobby store and a off the shelf microcomputer they can program.
@bjtaudio I designed and programmed a similar project back in high school in the 90s. The school didn't have facilities for manufacturing multi-layer boards so I built it single-layer and submitted the multi-layer design. That was an A.
If I'd submitted something like this during my engineering degree it would not have got a passing grade.
Later, as a postgrad teacher, if this had been submitted by a first-year student I'd have declined to mark it, but instead have invited the student to review their design process with me.
Engineering standards are a little higher than you seem to believe :D
There are some nice one out there for less than $2. quarter of the size of this one, that work for cooling or heating.
i guess it's kinda indicative of how readily available pcb printing and other related technology is over there- even people with basically no idea what they're doing can create circuits / products. it's cool in a way, while also scarey.
I've been thinking about making a temperature controller so thanks for the advice on what not to do.
First idea, when I saw it was, that the temperature sensor was wrong polarized. Trying a potentiometer instead of the sensor will perhaps bring a little light to what exact the circuit-board is doing. Sorry for bad English.I'm from Germany.
Uwe Soboszcyk That's how I would go about. Your English is just fine!
Connecting a potentiometer in place of K-type thermocouple does nothing, as the thermocouple has no resistance whatsoever, it generates a little voltage instead.
So what are the wires, superconductors?
Of course it has resistance. Just because you think it's low, doesn't mean it 'has no resistance whatsoever'.
Monkeh, if you want to electrically measure the wire length, you may ponder the resistance of a K-type thermocouple. If you want to measure temperature, as intended, the deciding parameter in a thermistor is its resistance, and as opposed to that, in a thermocouple, it's the induced voltage.
And yeah it has no resistance whatsoever. K-type wire often has less resistance than the sloppy Chinese screw terminal it connects through.
I'm well aware how a thermocouple works. That doesn't change the fact that the wire it is composed of has resistance.
Is it a good idea? NO - Will Clive try it? YES
I think I have a new phrase for dissing: "That's so modular!" Works in any situation.
Why is the left-most digit flickering a lot more than the others?
At first I thought it might be some sort of synchronized rolling shutter effect (temporal aliasing) with the camera, but it did the exact same thing after he power cycled the board as well.
If the entire display is being driven from the same chip, and even using the same single row, the only thing that I can think of is that there's something going on with the column driver corresponding to the leftmost digit.
its liberal , its trying not to be what it is , its non binary pansexual and kink
NOBOX7 5 internets for you.
I'm pretty sure that's happening, because the pins are shared for driving the 7-segment displays and reading buttons. One of the digits has to be switched to "input". Since a dedicated driver/button reader shouldn't be that bad and almost fail at what it's supposed to do (I mean, giving a useable display output is it's purpose), my guess is that whoever wrote the program read the buttons, but didn't set it back to output right after that.
superdau; Oh yes, bad programming does indeed explain it!
Thanks for pointing it out to me. (I'm more into the hardware design of things myself)
superdau I don't think the TM1650 allows the microcontroller to set the I/O state of the 7-segment pins directly. When reading which key is pressed, the microcontroller just sends a byte to the SDA/SCL pins and then reads a byte back. The TM1650 is in control of the multiplexing between keypad polling and 7segment led illumination.
Yes, Clive, you performed a fair test of this module. I would be ashamed to offer this "product" for sale when a minute or two of cursory testing shows it to be so flawed and in need of further development. Then again, I'm not Chinese.
Those LED segment displays are hard to see on camera. I figured out a little tip for that, especially the bright ones like the blue ones. When you peel off that protective shield it sometimes makes it brighter. Take a little strip of thin Kapton tape and place it in the segment display and it makes a big difference in cameras. Makes a huge difference if it’s a really bright one and easier to read on camera and in person. Hole this helps.
Recommended contrast materials include: spray glue and silvery PCB antistatic bag material, and window tint film.
Bog standard red electrical insulating tape works too, and is more likely to be in your toolbox.
A sure fix to that is reducing shutter speed a tiny bit
I use common red electrical tape on all of my displays. It works like magic. I can read them in full sun too.
PirateKitty
Doesn't that make your phone harder to see?
I’ve seen that non-shorting issue in a few cheap terminals...
i think the screw head isn’t part of the electrical design.. only the actual ‘plates’ that sandwich the terminal/spade.
If you’re unlucky, and the screw thread doesn’t touch the actual wire or spade - then it simply passes through the conductors to the captive nut on the other side of the mechanical assembly.
I can always get a connection at the mating surfaces... Odd, but there may have been a reason ??
LOL, I ordered a Temp control from BangGood just weeks ago. I had to insure this was not the one I had. I've used the MAX6675 with Arduino before. A Cap across the K ThermoCouple leads helps with noise. The MAX 6675 is a Current Amplifier. The K ThermoCouple generates a Current . They are immune to wire length and voltage loss.
Drinking game! Every time Clive says “modular” you have to take a drink.
I really dig double video day.
2:45 my guess: they used whatever their board design tool had for the standard pad size and just didn't think it through. i'm also kinda scared by the size of those pins, wouldn't use that terminal for any high power application to begin with, i feel like you probably want a bit larger diameter pins there. even something "low performance" like most barrel sockets have giant pins compared to that terminal strip.
My guess, the holes match the pins, but those clamps are unsuited to the needed currents.
did you try swapping the leads? that iron and constance sometimes get switched and or the silver solder junction goes bad
The MCU is a modern variation of the classic 8051 architecture. It might be worth it just as a devkit (if you have the necesary programmer. Bootleg ones go for $2 on ali.)
There is technically 3 connections to that thermal couple... if you look at the connections they are not part of the shielded lead, that shielded part needs to be grounded to the ground on the device. This is normally done when its attached to what ever your measuring via the chassi So if you add a wire to the shield and direct that to device ground, it will stop swinging
Another key thing is that thermocouples physically measure the difference between the temperature at the two wire ends, and the application needs to control the board end reference temperature to make sense. Scale may be in Fahrenheit, with a wrong reference for 32°F / °C, thus interpreting almost no difference as slightly below 30° nonsense.
2:39 I think the answer falls in line with your later conclusion that this is a sort of "text book" circuit design but with *many* errors. A sizable relay was used given the application, it was well bonded to the board, but the designer(s) utterly failed to realize (or care about?) the simple fact that the large loads this thing is supposed to be switching will be attached to those tiny terminals causing an obvious failure point. Only a hobbyist but that seems to make the most sense to me.
Today's drinking game. Take a drink every time Clive says "It is very very modular"
K type thermocouples often use Red as the negative output.... and often not isolated from the braid and mounting fixture. This may explain the 'noise' effect when handling and down register when the couple is heated.
Yeah... creep distance / clearance on the power supply and relay tracks is an issue... fix with cutting board between the tracks to effect air gap rather than tracking across a flat surface. BUT, you shouldn't need to do that.
Not sure if the testing methodology is correct for a thermocouple. For a PTC thermistor it would be but really if you shorted out a thermocouple you'd think it'd just display the ambient temperature. Because the connection to the thermocouple wire is in fact another junction at ambient temperature. Disconnecting the thermocouple just makes it read whatever the input is at if you leave it floating.
But what I want to know is 'is it modular' Clive?
It seems very modular, can you confirm that?
F
Yep
Nax FM module's nodules...IT'S MODULAR...only one thing for it..yes...GET ME THE SCIENTIFIC SCISSORS..hahaha
What if you connected the shield of the 'antenna' to ground?
Undefined Lastname do you mean the shield is connected to the blue connector (IN -)? With my comment I meant a permanent connection from DC ground to the shield. But maybe that is already the same.
Undefined Lastname Ok. Thanks :)
Did you try attaching a good thermocouple to it?
That thermocouple negative pin should be grounded RIGHT at the chip, instead they routed it threw the serial port, the voltages that thermocouples work on this will A) never hit 0, and B) will be very noisey
EDIT: it also looks like they intended a calibration via a pot to the analog pins but never fitted it
Others have pointed out that may be a position for a contrast pot and that the associated connector is for an LCD display.
Ah yes that makes much more since
Remember this temperature controller is modular, it's very modular, it's very very modular and it's a modular design that even the Borg would not assimilate into there collective even as a adjunct to there uni matrix zero one.
if you short the thermocouple terminals, I'd think it should display the ambient room temperature, as it would simulate 0 volts from the sensor. the way thermocouples work, they detect the temperature difference between the sensing end and the end connected to the terminals. there's normally a second temperature sensor on or near the board (preferably thermally coupled to the terminals) which detects the ambient temperature, then uses the thermocouple probe for fine adjustment.
if the probe end is warmer than the terminal end of the thermocouple, it'll feed a small positive voltage into the board. if the probe end is cooler than the terminal end, it will feed a small negative voltage into the board. it actually reads a relative temperature, not absolute.
That thermocouple chip needs a 100n capacitor between the thermocouple input pins. Debugged same issue before for another manufacturer...
Clive, you should try pulling the programming off the device.
I love watching these videos. Though more and more I'm realizing that Clive has probably forgotten more about electronics than I'll ever know in my life time. Keep the videos coming! Cheers.
You videos are always educative keep up the good work
Needs sending to Hydraulic Press Channel who will deal with it.
WineScrounger lol a very dangerous temperature controller that could attack at any time...!!
"What da fock?!?!"
do you mean AVE he will make it choochin!
Which hydraulic press channel?
anononomous ruclips.net/video/SnFYYNqwznY/видео.html
Accept no substitutes
Questions>>>> Wondering out of curiosity... One- The coupling of your hand on therm couple. Try a say 20 to 100 pico Farad cap across coupler input? Also wondering if put say 220 to 470 micro Farad across power input? See if that fix that issue. Far as some of the other things, relay going into bipolar disorder, no error on disconnect of probe so temp runs amok, these things I am laughing at myself. But was courious bout change in temp on that and if a few pico's on the them might help lots. Like the videos.
Its somewhat challenging to read that 7-segment display.. If to stick a piece of antistatic bag with a double sided tape numbers become easier to see.
Going full scale is a common open thermocouple indication on temperature controllers for heaters.
Aren’t there 50/60 suppression in these specialized ADC chips?
Same with debouncing in „keyboard controllers“.
Well, I've bought this for my DIY CNC Router to 3D Printer conversion. It was cheap and I thought it might work for some time. We'll see :D
It's arrived, but there are other components to come before I can start building the printer module.
No investigation as to WHY there's interference on the thermocouple input? The Max6675 datasheet says that the thermocouple negative should be grounded to the chip. Is it?
dear sir i have bought a 12volt crock pot from road pro worked great for about a month then i tried it didn't work the light came on no heat ruined my dinner.so i bought another one now the same thing lasted 2 weeks ruined another dinner contacted amazon they just said you bought it .i am not buying another i took it apart the wire is broken so i was just wandering what type of wire it goes to the element.it is a solid wire.thanks for your videos
Could it be the that one of the sides of the thermocouple chip isn't grounded?
Since the example schematic shows that one of the terminals should be grounded - which also explains how it starts reading about right when you ground the probe with your hand.
One side is connected to the negative rail. When I tried running it on a battery pack it was more stable,m so it doesn't like switching noise.
That looks to be quite modular !
If you were so inclined you could connect the microcontroller to a suitable reader and download the code. Then you could attempt to correct the omissions and re upload the code and see if it functions any better.
It would probably be a lot easier to work out the pins and rewrite the code from scratch.
bigclivedotcom
Seems logical enough.
stick a capacitor across those terminals, perhaps 100N ? see if that quietens it.
Take a shot every time you hear "modular" ;) Great video as always Clive!
I found a board pretty similar to this inside of a power supply for pool robots. It had the same type of relay.
Also the same finish
Hi, Clive have you checked that the thermocouple is connected the correct way round?
I didn't see you actually heat/cool the thermocouple, just play with it noise characteristic.
Have you tried a 0.1uF capacitor across the input terminals and one from one input terminal the gnd.
The capacitor on the PCB near the controller and the thermocouple chip is connected to what?
Tom..
It would be good to get your hands on a three term (PID) temperature controller
Big pads little pads! .... and I bet it came in a cardboard box.
Could you de-solder and remove all the on board power supply components (maybe pop in a 100nF and 100uF in parallel just for added noise removal )and just bridge over the input to the supply rails using your bench power supply set at the right voltage and see if the problem still persists? That would eliminate the on board power supply as a potential noise source? Could a suitable resistor across the probe input terminals also be used to stabilise/calibrate the probe? Use a different probe? Such a fun board with so many experiment possibilities :D
I would like to see some board like this, redone with as many of the original components as possible so that it actually works.
Hey Clive!
In my University in Germany we have Spartan-3E FPGA Boards. We use them to design our very own 4bit microprocessor from the ground up. I just thought it might give an interesting Video idea for how it might be used on all sorts of different prototyping
if shorting the sensor does not make a difference, I really start to wonder if that sensor is doing anything. Wonder what happens if you first ground it and then short the sensor. Maybe reading some screw-wire-screw thermocouple.
There is one thing good to say about this design though. I got the impression its very modular. Not sure what gave that away
Hey clive could you do a video where you muck about with some old 74xxx TTL logic chips? Binary counters, 7 segment displays, 555 timer.. I know a lot of that is redundant today with Arduinos and RaspRi's.. but it's still good stuff.
Here's a video I made about a slot machine I built with CMOS 4000 series logic.
ruclips.net/video/Q4aQiIXASuI/видео.html
You could have tested if the noise is from switching power supply by connecting your power supply straight to 5V.
I powered the module from a battery pack and it was more stable, but still inaccurate.
You could have guaranteed there would be noise coming from the switching power supply by connecting it straight to 240V.
At least momentarily, anyhow.
Lol I thought I was the only 1 that shorts out with scissors
I see you have a new magnifier and good because that old one was rather used up optically.
He just needed to refill the optical fluid
I think the pic has a resolution of 1024 bits. You just get a decimal readout of the anologue sensor's reading. Shame about the lack of ground / bad program.
If you want to use it for cooling, you just need to connect to the relay 's normally closed terminal. Right?
Paci. Exactly
We use these in the nuclear power plant I work at and have had no major disasters that you know about yet.
Have you tried powering it from a 12v battery to ensure your 12v power supply is not a source of noise in the circuit?
I did. And it resolved the ground coupled noise issue, but was still inaccurate.
Is it possible the solder pads are large on the relay to help conduct heat from the relay contacts as they age?
Hmm, you seem so sure. Guess you've never seen the solder connections melt on control boards as they age.
How modular is it?
To be fair the general rf noise floor of big Clive’s lab may only be matched by mr Carlson’s lab
I really like the relay disco mode, because it's such a stupid mistake anybody with any programming experience would have caught immediately.
If relayOff and temperature < TurnOnTemp then relayOn;
If relayOn and temperature > TurnOffTemp then relayOff;
It makes a good random Morse code generator.
Drinking game: have a shot every time Clive says modular.
Educational, as in "That'll teach ya!"
Is the resistance between the scissor prongs affecting the reading?
Are any of those big ground planes actually grounded, or are they just antennas?
Could it possibly be the solder connections where the thermocouple goes?
I have a different but similar module. But my p1, p2, etc. are not different presets but are functions such as number of degrees hysteresis. Below is another module different again. Are you sure of the functions on your one?
P0 Heat C/H C
P1 Backlash Set 0.1-15 2
P2 Upper Limit 110 110
P3 Lower Limit -50 -50
P4 Correction -7.0 ~ 7.0 0
P5 Delay Start Time 0-10 mins 0
P6 High Temperature Alarm 0-110 OFF
The take away for me, the sensor connector itself has a bad connection to the board, when you pushed down with the scissors it forced a connection, I'd say it needs to have the solder run.
Hmmm... why has this video only the left audio channel?
Check your surround sound settings. This has been happening on a lot of channels.
You should be able to get that to read right. I've found that thermocouple are quite good antennas, I think the dissimilar metals rectify noise and create an offset. You should be able to sort that
Ebay, the dumping ground for dodgy & dud electronics. That relay chatter is especially Dudley.
Suprised the shield on probe isn't grounded?
I wonder if the shield is actually the problem. I'd try a different 2 wire probe.
Would love to see you redesign and fix it
If the thermocouple has a different range than the unit is calibrated to, you'll see problems like you're seeing. Also, did you try running a line from the probe- to ground? while the probe is connected that is?
Big Clive, can you help me find the *good* dual temperature controller you reviewed a couple of months ago, please?
This one? ruclips.net/video/9zymx2xgTmk/видео.html
That's it!! thank you very much.
Shouldn't the two thermocouple leads be taken to the input independently of the screen braid which should be grounded. Even if one of the thermocouple leads is grounded at the chip?
Also could it be something as simple as a faulty decoupling capacitor?
How about another video improving the design?